Track initialisation for multiple formations based on neutrosophic Hough transform

Track crossing is a major issue in the initialisation of multiple-flight trajectories. To solve this problem, the authors propose a track initialisation algorithm. It uses characteristics of formation flight to find centroids of each formation with DBSCAN algorithm. Then, it initialises the track based on the centroid. The author use a Neutrosophic Hough Transform (NHT) method to improve accuracy and computational speed. That helps address errors caused by the approximation of straight lines between true points resulting from the clustering algorithm. The authors made three experiments using track initialisation data from two flight formations with five target aircraft each, over a span of three frames. NHT, Fuzzy HT, Improved Hough Transform (HT) and HT are compared. Results revealed that the average runtime of NHT was 10.2153 s. The F-measure of NHT was 100.00%, while that of Fuzzy HT was 9.8347 s. The F-measure of Fuzzy HT was 80.00%. The Improved HT was 12.0723 s. The F-measure of Improved HT was 11.76% and HT was 13.783 s. And the F-measure of HT was 6.87%. The authors lost some computation speed to achieve higher prediction accuracy. The accuracy of the NHT is higher than other methods.